Finding Dark Matter Faster with Explicit Profile Likelihoods

Liquid xenon time-projection chambers are the world's most sensitive detectors for a wide range of dark matter candidates. We show that the statistical analysis of their data can be improved by replacing detector response Monte Carlo simulations with an equivalent deterministic calculation. This allows the use of high-dimensional undiscretized models, yielding up to $\sim\! 2$ times better discrimination of the dominant backgrounds. In turn, this could significantly extend the physics reach of upcoming experiments such as XENONnT and LZ, and bring forward a potential $5 \sigma$ dark matter discovery by over a year.